Hi HN, We're Davide, Gianluca and Marco from INAYA. We’re building autonomous AI analysts that help manufacturing procurement plan and buy under commodity price volatility. We applied to YC W26 and are waiting for feedback. We previously led AI/Data programs at a Big Four with manufacturing clients; that vantage point made the procurement wedge obvious.

THE PROBLEM (AND WHY NOW)

Procurement is being asked to manage more risk with fewer people and tighter budgets. Price swings collide with slow approvals, expiring supplier quotes, and data scattered across tools. The result: missed hedging windows, re-sourcing under pressure, and margin leakage that finance sees only after month-end. Recent CPO surveys show risk management and resilience have moved to the top of the agenda; teams need earlier signals, clearer explanations, and faster decisions — volatility didn’t end; slack did. - Sunflower oil +39% (Q1 ’25) - Aluminum premiums +250% (Oct ’25) - Cocoa +85% (Apr ’25)

WHAT WE BUILT FIRST (BEFORE AGENTS)

To validate the need, we shipped four apps that teams use today:

1) Global OSINT Engine Configurable and fully parametric. Define query templates, select countries and languages, and capture local-language signals (policy, logistics, trade) across 80+ languages and 230+ countries. It often surfaces indicators before market feeds.

2) Predictive Core Uses market, macro, and sentiment features (500+ signals) to generate price forecasts up to 18 months. On commodities currently released in our platform, backtests show 90–95% accuracy, outperforming the tools our pilot manufacturers use today. (We can share evaluation details—splits, metrics, baselines—in the comments.)

3) Spot & Forward Prices Combines quoted and non-quoted spot prices with forward curves from major exchanges to give a full time view and a simple way to compare forecasts to market structure.

4) Multimodal What-If Runs instant simulations to quantify exposure and show how price shifts flow into unit costs and margins (useful for make/buy timing and hedging).

Together, these apps give procurement comparable views (signals ↔ forecasts ↔ forward curves ↔ P&L impact) so decisions aren’t made in the dark or too late.

WHAT WE'RE ADDING ON

We’re adding an agent that uses these apps together, so teams can work conversationally instead of hunting through dashboards.

Ask:

- “Why did copper rise this week?” - “What happens to our cost base if gas is +10%?” - “Simulate Chinese export limits on copper.”

The agent runs OSINT → Forecasts → Spot/Forward → What-If, cross-checks the results, and returns a sourced explanation in plain language (in the app or via Teams/Slack). The goal isn’t to replace judgment but to shorten the path to a defensible decision.

WHERE WE ARE

Early access with manufacturers in the EU and US. Next milestones: proactive alerts, deeper ERP/P2P integrations, and public notebooks for long-horizon forecast evaluation.

WHAT WE’D LIKE FEEDBACK ON

- Better ways to evaluate long-horizon forecast quality (beyond MAPE/hit-rate). - How to design explanations users can trust (data traces, competing hypotheses, ablations). - First integration points you’d want (ERP, P2P, Teams/Slack workflows). - Any additional valuable feedback is more than welcome.

LINK

Website: https://www.inaya.ai

Thanks for reading — happy to go deep on the technical details.

.NET 10

Discussing Beyond Vibe Coding by Addy Osmani; Embracing AI as a Senior Engineer [video]

Show HN: Run parallel agents in VSCode tabs

Generative AI for Automotive Critical Scenario Generation

MoM – Mixture of Model Service

Backblaze Drive Stats for Q3 2025

SAP fixes hardcoded credentials flaw in SQL Anywhere Monitor

Google claims Microsoft 'bribes' people to use Edge

Show HN: I just wanted to fix my heating but built a Cursor agent instead

Hosting Minecraft on a cheap AliExpress smart lightbulb

Windows 95 application compatibility patched broken programs

Ask HN: Do you have any SSH bot scripts?

A tapestry woven from logic gates and advent-blinking lights – a Number Garden

Show HN: SpecMind – AI architecture tool for vibe coding

Ask HN: Has AI unlocked probability proofs for computer programmers?

Former Trump Official to Lead Israeli Spyware Maker NSO Group

Ofcom is monitoring VPNs following Online Safety Act

Paramount Plus is getting a price hike in early 2026

LLM-Driven Robots Risk Enacting Discrimination, Violence, and Unlawful Actions

Bandits in Your LLM Gateway

Ask HN: Senior people, how did your career evolve?

Evaluating urban methane emissions and their attributes in a Osaka, Japan

RAM prices are out of control

Tired of being manipulated by BigTech?

Show HN: Inaya – AI agent for manufacturers managing commodity volatility

Show HN: PocketWise – a simple, natural-language double-entry finance tracker

Lfg.gg – Looking for Games

Mathematician solves one of the most famous unsolved conjectures in Math

Proving Safety at Scale: Spark, RISC-V, and Nvidia's Security Strategy

Nimony v0.2 – early preview of Nim 3.0's compiler

Show HN: Self host your apps, domains, SSL. Vercel UX on your VPS